J. Phys. Chem. B 2001, 105, 2393-2403
2393
Adsorption of Molybdate Monomers and Polymers on Titania with a Multisite Approach K. Bourikas,† T. Hiemstra,* and W. H. Van Riemsdijk Department of EnVironmental Sciences, Wageningen UniVersity, P.O. Box 8005, NL 6700 EC Wageningen, The Netherlands ReceiVed: June 22, 2000; In Final Form: NoVember 8, 2000
Adsorption of polymers on mineral surfaces is of great interest. A representative important system is the molybdenum supported titania, used in catalysis. The well-known chemistry of molybdate polymers in aqueous solutions allows a detailed study of the contribution of these polymers to the Mo adsorption on titanium oxide surface. A large range of Mo concentrations was covered: from low levels where only Mo monomers are present to high levels where extensive polymerization takes place. Data from different techniques, like potentiometric titrations, proton-ion titrations, and adsorption edges, were used for the description of the interface “molybdate solution/titania surface”. All the data could be modeled very well using the recently introduced CD-MUSIC approach. The charge of the surface complexes is spatially distributed in the interface. MoO42- monomers adsorb over the entire pH range forming inner sphere complexes, which are characterized by two types of structure, i.e., bidentate and monodentate. Below pH 5.5 and high total molybdate concentrations the heptamolybdate polyanion Mo7O23(OH)5- is also adsorbed, forming an outer sphere complex. Due to its relatively large size, it covers a significant number of surface groups.
Introduction Anion and cation adsorption at the solid/solution interface of metal (hydr)oxides is an important phenomenon in many fields of chemistry. It is of great interest among soil chemists and geochemists studying the natural environment, among colloid chemists studying colloidal systems and many scientists involved in chemical engineering. Over the past decades, there also have been an increasing number of related studies in catalysis. A significant part of heterogeneous catalysis is dedicated to the supported catalysts.1,2 These materials are usually prepared by impregnating the supportswhich is usually a metal oxides in an electrolytic solution containing the active element to be deposited. Filtration, drying, and calcination follow. The impregnation step, where adsorption of the active element on the oxide surface takes place, is often critical because it can significantly determine the properties of the final catalyst. Depending on the impregnating conditions (pH, concentration, ionic strength, etc.), several ionic species of the element may occur in the suspension. For several elements, high concentrations result in the formation of polymers in the suspension. The polymers may be adsorbed on the support surface. Elucidation of the factors determining the adsorption process of the polymers will allow the tailoring of the preparation of the catalyst by regulating the impregnation parameters. Various models have been used to describe adsorption data.3-12 They are tested on the basis of the possibility of describing these data and predicting results. Most popular are the surface complexation models including an electrostatic double layer option. For a limited set of data many different models give similar results.13 As a result of this, there is an * To whom correspondence should be addressed. E-mail:
[email protected]. † On leave from the Department of Chemistry, University of Patras, and the Institute of Chemical Engineering and High-Temperature Chemical Processes, ICE/HT-FORTH, P.O. Box 1414, GR-26500 Patras, Greece.
increasing demand for new experiments and new information from other related fields that could help in the direction of discriminating among all these models. Particularly, the use of in situ spectroscopic techniques, revealing the physical chemical structure of the surface species, has provided useful information for the models.14-22 In most commonly used surface complexation models it is difficult to take advantage of structural details since species are treated as point charges. This difficulty has been overcome with the CD-MUSIC model (charge distribution-multisite complexation).23 This model emphasizes the importance of the surface structure and the structure of adsorbed species. Surface complexes are not treated as point charges but are considered to have a spatial distribution of charge in the interfacial region. The model has been applied in several oxy-anion systems related to adsorption of monomers.23-28 In most cases, its predictions are consistent with physically realistic surface complexes found by spectroscopy. In this paper, we will apply this model to describe the adsorption of molybdenum on the surface of titanium oxide. Our approach is based on a large range of adsorption data, from very low to high adsorption levels, where polymeric molybdate species are formed. This system was selected for two important reasons: The first is its great importance and use in the heterogeneous catalysis. The second is the simple and wellknown chemistry of the aqueous molybdate solution, even at high concentrations where polymers are found. The study of this system may show the way that could be followed for studying similar important systems, where polymers are included in an adsorption process. Studies dealing with supported molybdate catalysts prepared by the adsorption method are not abundant in the literature.29-33 Most of these studies focused on the structure of the adsorbed molybdates after calcination of the samples. Although this kind of characterization can give useful information about the properties of the final catalyst, it cannot reveal the adsorption
10.1021/jp002267q CCC: $20.00 © 2001 American Chemical Society Published on Web 03/03/2001
2394 J. Phys. Chem. B, Vol. 105, No. 12, 2001 mechanism of the molybdates on the titania surface. The first detailed study to this direction was done by the group of Lycourghiotis.34,35 Using an approach11 based on the widely known homogeneous two pK - triple layer model, they described a wide range of adsorption data quite well. However, the final surface speciation remains uncertain due to a number of simplifying assumptions. Particularly, the treatment of the adsorbed species, even the large polymers, as point charges and the nonrealistic picture of a homogeneous surface as well as the relatively large amount of the adjustable parameters are a part of that. In this paper, by applying the CD-MUSIC model and using the experience of the previous approach, we will try to overcome these limitations, improving our understanding of the nature and the structure of the adsorbed molybdates on titania. To separate clearly between monomer and polymer adsorption, we will first study the monomer adsorption extensively. This can be done at low levels of molybdate loading. At these conditions, a very powerful tool for characterizing accurately the adsorption is the determination of the proton-molybdate stoichiometry at constant pH. This will be done my measuring the proton coadsorption as a function of the molybdate adsorption. It provides direct information about the pH dependency of the adsorption.36,37 It is also useful in the interpretation of the binding structure of the adsorbed molybdate.38 Particularly, a sensitive determination of the charge distribution of the adsorbed species can be achieved. After characterization of the monomer adsorption, the model will be applied to conditions where polymeric molybdate species are adsorbed. Materials and Methods Synthesis and Characterization. All chemicals (Merck p.a.) were stored in plastic bottles, and all experiments were done in plastic vessels to avoid silica contamination. The temperature was kept constant at 25 °C, and the ionic strength was equal to 0.1 M. Ultrapure boiled water was used to prepare all solutions. For the proton-molybdate titrations and the adsorption edges at low levels, pure anatase was used, prepared by hydrolysis of titanium isopropoxide (Alfa).39 The specific surface area (SSA), determined with BET measurements, is 40 m2 g-1. Sodium molybdate and sodium nitrate stock solutions were prepared from the corresponding crystalline reagents. For the edges at high adsorption levels, commercially available titania (Degussa P25) was used, with a specific surface area of 50 m2 g-1. The patchwise structure of the surface consists mainly of anatase.40-42 Ammonium heptamolybdate and ammonium nitrate stock solutions were prepared from the corresponding crystalline reagents. The choice of the commercial titania as well as of the ammonium heptamolybdate and nitrate was for reasons of comparing results with previously reported data on the same system.34 Moreover, this titania is the most commonly used in the preparation of titania supported catalysts. The use of ammonium reagents is preferred in catalysis because of the easy removal of ammonium in the heating process. In the case of low adsorption levels, we used sodium molybdate and nitrate instead of ammonium heptamolybdate and nitrate in order to avoid possible proton release due to transformation of NH4+ to NH3 at high-pH values. This could influence the protonmolybdate coadsorption stoichiometry. Mo Speciation in Solution. Ammonium heptamolybdate solution (0.016 25 M) was added to 50 mL of 0.100 M NH4NO3 solution at 25.0 °C in a vessel purged with moist N2 gas,
Bourikas et al. and the pH was registered. The experiment was repeated with 0.001 625 M (NH4)6Mo7O24 in 0.090 M NH4NO3 added to 49 mL of 0.100 M NH4NO3. The 0.100 M NH4NO3 solution was slightly acidic, having an initial pH of the 5.32, probably due to removal of very small amounts of NH3(g). In titration experiments 0.375 or 3.75 mL of 0.016 25 M (NH4)6Mo7O24 was added to 75 mL of 0.100 M NH4NO3 and titrated with 0.0976 M NaOH. A N2 atmosphere was kept above this solution (25.0 °C). Proton-Ion Titrations at Constant pH. Vessels containing approximately 50 mL of 20-60 g L-1 titania were purged with moist nitrogen gas at pH < 6 for 1 day to remove CO2 (I ) 0.1 M). Next, the pH of the suspension reached the appropriate pH of the experiment with the addition of a small amount of acid (HNO3) or base (NaOH). The suspension was kept (pH-stat) at this pH for about 1 h to reach equilibrium. Then, it was titrated with a 0.01 M sodium molybdate solution. During each addition of 0.2-0.5 mL of Na2MoO4 the pH was kept constant by titration with standardized 0.01 M HNO3. A reaction time of about 20 min was enough for the system to reach equilibrium. Under the experimental conditions the solution Mo concentration was negligible compared to the adsorbed one, giving practically almost 100% adsorption (except pH ) 8). This was certified by taking a filtrated sample for analysis with ICP-MS after the end of each titration. The proton coadsorption is calculated from the difference between the number of protons added to the suspension and the very small change in the number of protons in the solution. If almost no Mo remains in the solution, the Mo adsorption follows directly from the added amount of Mo.38 Adsorption Edges. Adsorption experiments were done at different initial Mo concentrations (10-6, 10-5, 10-4, 10-3, 10-2, and 2 × 10-2 M). They were performed in individual plastic tubes (20 mL), with fixed amounts of titania, electrolyte, molybdate, and different amounts of acid or base, to give adsorption edges. To avoid contact with CO2 at high-pH values, a CO2 free titania suspension was mixed with the solution in the tubes, under nitrogen atmosphere. The tubes were closed immediately and then equilibrated for 24 h by end-over-end rotation. After measuring the final pH, the tubes were centrifuged and samples of the supernatants were taken for analysis with ICP-MS. The amount of the adsorbed molybdate was calculated from the difference between the total initial Mo concentration and that of the supernatant. Model Calculations. Calculations were carried out with ECOSAT,43 a computer code for the calculation of chemical equilibria. The solution and the surface equilibria used are presented in the Appendix. Model Description Multisite Complexation (MUSIC). Metal oxides are characterized by the presence of different types of surface groups. The multisite complexation model takes into account this surface heterogeneity. Variation in types of surface oxygens is due to a different number of coordinating metal ions. The charge on these surface oxygens can be found by applying the Pauling bond valence concept.44 In this concept, the charge of the central metal ion is distributed over the coordinating oxygens. The structure of titanium oxide consists of Ti4+ filled oxygen octahedra. The oxygens in the bulk of the solid are triply coordinated (Ti3O0), receiving from each Ti4+ a bond valence of 2/3 (each Ti4+ ion distributes its charge over six surrounding oxygens). The surface oxygens may be singly, doubly, and triply coordinated, depending on the number of the coordinating Ti4+ ions. Their coordination is lower than those of the bulk because
Mo(VI) Monomer and Polymer Adsorption on Titania
J. Phys. Chem. B, Vol. 105, No. 12, 2001 2395
of some missing bonds. The binding of one or two protons can compensate the missing charge. In the case of singly coordinated groups, three species can be defined: TiO-4/3, TiOH-1/3, and TiOH2+2/3. It has been shown45-47 that the constants of two consecutive protonation steps differ very strongly. For the singly coordinated Ti groups this implies that only the second protonation step is of relevance for the charging behavior. On the basis of the above concepts, the charging behavior of the titania surface can be described by the following protonation reactions of singly and doubly coordinated groups: KH1
TiOH-1/3 + H+ 798 TiOH2+2/3 KH2
Ti2O-2/3 + H+ 798 Ti2OH+1/3
(1) (2)
Protonation of the uncharged triply coordinated groups, Ti3O0, is not possible in the normal pH range. Therefore, these groups are considered inert in normal conditions. An extended description of the MUSIC model can be found elsewhere.45,46 CD Approach. An extension of the above-mentioned bond valence concept to surface complexation forms the basis of the CD model.23 The charge of surface complexes is distributed over two electrostatic planes. A schematic representation of this idea is given in Figure 1. The charge of the surface groups is located at the surface plane (0-plane). Adsorbed ions are distinguished in inner sphere complexes and outer sphere complexes. The inner sphere complexes are assumed to have a spatial distribution of the charge. Part of their charge is attributed to the surface (since not all ligands of the adsorbed complex are common with the surface) and the remaining of the part is attributed to the first plane (1-plane), at a certain distance from the surface. This means that a fraction f of the charge of the central ion (zMe) of the complex is placed at the 0-plane and the remaining fraction (1 - f) is placed at the 1-plane. The change in charge in the 0-plane (∆z0), due to the ion adsorption, is determined by the fraction of the charge of the central ion placed on the surface (fzMe) plus the change of the charge due to the change in the number of protons (∆nH) on the surface ligands involved in the surface reaction. Thus, ∆z0 can be defined as
∆z0 ) ∆nHzH + fzMe
(3)
where zH is the valence of the proton. The ∆nH depends on the number of common ligands and the change in the state of protonation of these ligands. Its value may be positive, zero, or negative. The change in charge in the 1-plane (∆z1) is determined by the fraction of the charge of the central ion placed in the 1-plane, (1 - f)zMe, plus the sum of the charges of the solution oriented ligands (∑mjzj):
∆z1 ) (1 - f)zMe +
∑j mjzj
(4)
where mj and zj are respectively the number and the charge of these ligands. In case of an ion surrounded by four ligands, as presented in Figure 1, an equal distribution of the charge of the central ion gives f ) 0.25 for a monodentate complex (since one of its four ligands is common with the surface) and f ) 0.5 for a bidentate complex (since two of its four ligands are common with the surface). The fraction f is also called the CD value. It should be mentioned here that, as has been suggested,48-50 only the singly coordinated surface groups are reactive for inner
Figure 1. Schematic representation of the metal (hydr)oxide/solution interface (TP). Surface groups are coordinated with metal ions of the solid phase, and the corresponding charge is located in an electrostatic plane (0-plane). The surface groups may form inner sphere complexes with adsorbed ions. The surface-oriented ligands of inner sphere complexes are also located in the surface plane (0-plane). A bidentate surface complex has two common ligands, a monodentate surface complex one. The solution directed ligands of the inner sphere complexes are located in the inner plane of the compact part of the double layer (1-plane). The charge of the central ion of the inner sphere complexes is distributed over both electrostatic planes. Pair forming ions are treated as point charges and placed in the outer plane (2-plane). The space between a set of electrostatic planes is characterized by a capacitance.
sphere complex formation of the anions. Considering an inner sphere molybdate complex binding to a singly, doubly, or triply coordinated group, we can calculate the sum of the bond valences on surface oxygen (Ti-O-Mo, Ti2-O-Mo, Ti3-OMo). Assuming a Pauling bond valence for a Mo-O bond of 1.5 vu (valence unit) and for a Ti-O bond of 2/3, this sum is equal to 0.17, 0.83, and 1.5 vu, respectively. According to literature,51 only a neutral or almost neutral sum of bond valences is possible, implying that only the singly coordinated surface oxygens are reactive for an inner sphere molybdate complexation. The outer sphere complexes of electrolyte ions are placed at a position determined by the minimum distance of approach of hydrated ions to the (hydr)oxide surface. They form ion pairs with the surface groups without having common ligands with them. They are located in the outer plane (see Figure 1) and treated as point charges in the CD model, like the counterions of the diffuse part of the double layer (DDL).
2396 J. Phys. Chem. B, Vol. 105, No. 12, 2001
Bourikas et al.
In the case of an absence of surface complexation, the model becomes identical with the basic Stern layer model, and the compact part of the double layer is characterized by the wellknown Stern capacitance (C). Inner sphere surface complexes divide the Stern layer in two parts, the inner layer with C1 capacitance and the outer one with C2 capacitance. These are related to the overall Stern layer capacitance according to the following:
1 1 1 + ) C C1 C2
(5)
The above-presented electrostatic approach is known as the three plane model (TP).13,23 The TP differs from the widely used triple layer model (TLM) on the following parts: (i) In the TLM, the adsorbed species are considered as point charges usually located at the surface plane. In the CD-MUSIC model, the adsorbed species are assumed to have a spatial distribution of their charge over two electrostatic planes (0 and 1 planes). (ii) In the TLM ion pair formation is placed in the 1-plane, while in the TP model ion pairs are located in the 2-plane. (iii) The capacitance of the outer layer in the TLM is fixed at a value of 0.2 F m-2,52,53 in an effort to link the double layer properties of AgI and Me (hydr)oxides. It has been pointed out that the linkage can be interpreted differently,54 leading to much higher outer capacitances of the double layer in metal (hydr)oxides. In the CD-MUSIC approach, a value of 5 F m-2 for C2 has been suggested.23-28 Results and Discussion Molybdate Solution Chemistry. The formal oxidation state of molybdenum in aqueous solutions is +6 over a broad potential vs pH region. Various oxy-anions are formed whose presence and relative amount depends on the pH as well as on the total molybdate concentration.55-57 At high-pH values (above 6), Mo is present as the tetrahedral monomeric molybdate ion MoO42-. This can be protonated at low-pH values forming the H2MoO4. According to Baes and Mesmer,55 the presence of HMoO4- is uncertain. Decreasing the pH at concentrations above 10-4 M leads to the formation of the heptamolybdate polyanion, Mo7O246-, also known as paramolybdate, with a structure that contains MoO6 octahedra. Depending on the acidity in solution, the heptamolybdate can be singly, doubly, or even triply protonated. At very low-pH values (below the pH range of interest), other types of Mo polymers are formed. For the conditions of this study the solution chemistry can be described by the following equilibria: K1
MoO42- + H+ 798 HMoO4K2
MoO42- + 2H+ 798 H2MoO4 K3
7MoO42- + 8H+ 798 Mo7O246- + 4H2O K4
7MoO42- + 9H+ 798 Mo7O23(OH)5- + 4H2O
(6) (7) (8) (9)
K5
7MoO42- + 10H+ 798 Mo7O22(OH)24- + 4H2O (10) The published values of the above-mentioned equilibrium constants have been mainly determined at very high salt levels.55,58 However, our experiments as well as those of other groups have been done at 0.1 M ionic strength. Due to the
Figure 2. Variation of pH with total molybdate concentration (a,b) added as (NH4)6Mo7O24 at 25 °C and I ) 0.1 M NH4NO3. Titration of two ammonium heptamolybdate solutions with NaOH at 25 °C in 0.1 M NH4+ + Na+ (c). The triangles in b represent experimental data obtained from refs 34 and 59. Solid lines correspond to the calculated curve based on the equilibria of Table 1.
considerable influence of the high charge of polymers on the activity coefficient, we needed new values for the equilibrium constants referring to 0.1 M ionic strength. To calculate these values, we fitted our experimental data (T ) 25 °C), using equilibria 6-10 (Figure 2). We have calculated the log K1 value for I ) 0.1 M from the value in 1 M NaCl, since our data are not particularly sensitive for the log K1 value. The other log K values were fitted. The constants K1-K5 are presented in Table 1, together with the values obtained from ref 55. In the calculations we used activity coefficients estimated with the Davies equation (constant ) 0.2). The speciation in the solution, based on the calculated values for the equilibrium constants, is presented in Figure 3. It is noticeable that above pH 6 only the monomeric MoO42- ions are present in the solution at our highest Mo concentrations (≈0.01 M). At lower pH values and high total Mo concentrations (>10-3 M), the polymers Mo7O246-, Mo7O23(OH)5-, and Mo7O22(OH)24- are formed. On the basis of our formation constants of HMoO4 and H2MoO4, we predict the dominance of MoO42- and H2MoO4 as monomeric species at low concentrations
